Transcript PowerPoint Slides - University of Georgia

Welcome
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
PROLOG/RDBMS Integration
In The NED Intelligent
Information System
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
Participants
University of Georgia
F. Maier
D. Nute
W.D. Potter
J. Wang
M. Dass
H. Uchiyama
USDA Forest Service
M. J. Twery
H. M. Rauscher
P. Knopp
S. Thomasma,
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
NED Goal: provide a set of tools for
Natural Resource Decision Support
• NED provides
- a set of Decision-Support Tools
- analysis for integrated prescriptions
- multi-variable forest management
- multi-scale support from plot to landscape
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
The NED Decision Process
1.
2.
3.
4.
5.
6.
7.
Create the goals & measurement criteria
Inventory & current condition analysis
Design alternative courses of action
Forecast the future through simulation
Assign values to the measurement criteria
Evaluate how well goals have been met
If not satisfactory, go back to step 1
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
NED Architecture
Knowledge Models
Inference Engines
Meta-knowledge
Blackboard
Temporary Data Files
Simulators
GIS
Visual Models
HTML
Report
s
Interface Modules
A
G
E
N
T
S
Prolog
Clauses
MS Access
Databases
Control Flow
Information Flow
Heterogeneous Sources
As an Intelligent Information System, NED
provides seamless integration of (possibly
heterogeneous, distributed):
• Microsoft Access Databases (e.g. inventory)
• Knowledge Bases (e.g. treatments and goals)
• Simulation Sources (e.g. FVS and Silvah)
• Visualization Sources (e.g. Arcview and
Envision)
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
PROLOG/RDBMS
The ProData method to query a database did
not meet the needs of NED-2 because :
• Processing data from multiple tables is slow
• It requires the database schema to be known
• Changes to the database schema are allowed
only at design time (not during operation)
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
Integration Techniques
Techniques for Integration (Brodie & Jarke, 1988):
• Coupling existing PROLOG & RDBMS
implementations
• Extending PROLOG to include DBMS
• Extending DBMS to include PROLOG
• Tightly integrating LP techniques with
DBMS techniques
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
NED-2 Query Process
Initiator
B
L
A
C
K
B
O
A
R
D
Database
Meta
Data
NED-2 Agents
ProData
ODBC
Database
Northeastern Research Station
Southern Research Station
Database
Database
Database
The University of Georgia
Artificial Intelligence Center
NED-2 Feature
Special Feature in NED-2
Ability to retrieve information from multiple
data sources without having to specify,
within a query, where the data is to be found
(e.g., in DBs, KBs, or as the result of
simulations). Metadata is the key.
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
CREATING METADATA
Creating metadata dynamically..
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
Query Example
Query : What is the area of the stand in
snap shot 0 ?
Prolog Query :
known(‘STAND_AREA’([‘SNAPSHOT’ = 0])).
Query: ‘STAND_AREA’ = X, ‘SNAPSHOT’ = 0
Source Matching :‘STAND_HEADER’:
‘STAND_AREA’ = X,
‘SNAPSHOT_TREATMENTS’: ‘SNAPSHOT’=0
Join Constraints :‘STAND_HEADER’:‘STAND’ =
‘SNAPSHOT_TREATMENTS’:‘STAND’
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
Query Example (cont.)
SQL Statement :
SELECT ‘STAND-HEADER’.‘STAND-AREA’
FROM
‘STAND-HEADER’
‘SNAPSHOT-TREATMENTS’
WHERE
‘SNAPSHOT-TREATMENTS’.‘SNAPSHOT’ =
0
AND
‘STAND-HEADER’.‘STAND’ =
‘SNAPSHOT-TREATMENTS’.‘STAND’
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
Query Language Features
•
•
•
•
•
•
Arithmetic operations
Logical Operations
Aggregates
Subqueries
IN and BETWEEN
DISTINCT and ALL
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
Conclusion
• Makes full use of database capabilities.
• Faster query set-up and processing.
• No need for full knowledge of a schema.
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center
Further Information
http://www.fs.fed.us/ne/burlington/ned/
Northeastern Research Station
Southern Research Station
The University of Georgia
Artificial Intelligence Center